Download Nat. Struct. Biol. 8, 192-194.

© 2001 Nature Publishing Group http://structbio.nature.com
news and views
Two insulators are not better than one
© 2001 Nature Publishing Group http://structbio.nature.com
Fabien Mongelard and Victor G. Corces
Chromatin insulators seem to control eukaryotic gene expression by regulating interactions between enhancers
and promoters. Mounting evidence suggests that these sequences play a complex role in the regulation of
transcription, perhaps mediated by changes in chromatin structure and nuclear organization.
The expression of eukaryotic genes in specific tissues and at particular times of
development is determined by regulatory
sequences called enhancers and silencers.
Enhancers bind transcription factors,
which then help initiate transcription of
the RNA at the promoter. Enhancers
accomplish this independently of their
distance and location with respect to the
gene. This flexibility in enhancer function
also gives these sequences the potential of
being highly promiscuous. Since this
promiscuity is rarely observed, eukaryotes
must have developed mechanisms to
insure that genes are not activated in the
wrong tissue or at the wrong time of
development by enhancers from a neighboring gene. Chromatin insulators may
play the role of gatekeepers that curtail
enhancer promiscuity.
Boundaries or insulator elements have
been experimentally identified based on
two defining characteristics. Insulators
interfere with interactions between
enhancers and promoters and inhibit
enhancer-activated transcription, specifically when interposed between the affected enhancer and the promoter. Insulators
also buffer transgenes, inserted in the
genome at random sites, from chromosomal position effects1. Based on the first
property, insulators can be viewed as an
additional class of regulatory sequences in
the arsenal of controlling elements at the
service of genes to ensure their proper
spatial and temporal regulation. But the
latter property seems to suggest a more
complex role, perhaps involving the organization and compartmentalization of the
chromatin fiber within the nucleus. Two
papers recently published in Science
appear to support this interpretation2,3.
Properties of an insulator
Insulator elements have been described in
many organisms, including yeast,
Drosophila and vertebrates1. In yeast, insulators appear to delimit the boundaries of
silenced chromatin at telomeres and the
mating-type loci HML and HMR.
Drosophila insulators were the first to be
characterized and this organism displays
192
the largest collection of these sequences so
far. The specialized chromatin structure
(scs and scs′ elements) flank one of the
heat shock protein 70 (hsp70) gene loci,
where they seem to delimit the region that
becomes transcriptionally active (or
puffed) in response to heat shock. Other
Drosophila insulators include those found
in the so-called gypsy retrovirus and the
Fab-7 and Fab-8 elements of the bithorax
complex, a complex containing three of
the Hox genes that encode important
developmental control proteins in flies.
Insulators have also been studied in vertebrates, where they have been found in the
ribosomal RNA genes of Xenopus, the
chicken β-globin genes and the human T
cell receptor (TCR)-α/δ locus among others. The widespread distribution of insulators is suggestive of an important role
for these sequences in the biology of the
nucleus.
An idea of the role insulators could play
in the cell can be gleaned from the analysis
of their distribution in the bithorax complex of Drosophila. The various genes of
this complex are expressed in a parasegment-specific pattern (a parasegment is a
metameric unit along the anterior-posterior axis of flies that is out of phase with segments) by a complex set of regulatory
sequences arranged in a linear fashion over
300 kb of DNA, corresponding to the order
of expression of the genes along the anterior-posterior axis of the animal. These
parasegment-specific regulatory sequences
appear to be separated by insulator elements, one of which is called Fab-7. The
Fab-7 insulator is located between two
enhancers named iab-6 and iab-7 that control expression of the Abdominal-B
(Abd-B) gene in parasegments PS11 and
PS12, respectively. Deletion of the Fab-7
insulator eliminates the gatekeeper normally ensuring that the iab-6 and iab-7
enhancers activate transcription of the
Abd-B gene in the appropriate tissues.
Without the Fab-7 insulator, the iab-7
enhancer activates expression of Abd-B in
the wrong cells, causing homeotic phenotypes in the adult fly. Homeotic phenotypes are those in which homologous
structures, such as legs and antennae, are
changed into one another. These results
suggest that the Fab-7 insulator is required
for proper gene expression in order to
avoid interference among the different regulatory sequences of the Abd-B gene4–6. A
second example that illustrates the role of
insulators in vivo is that of imprinting in
mice. Imprinting is an epigenetic trait
often seen as differences between paternal
and maternal inheritance. The parent-oforigin-specific expression of the mouse
H19 and Igf2 genes is controlled by an
insulator element located between the two
genes7,8.
Two possible models
Much of the work on chromatin insulators
has centered on the characterization of
associated proteins with the goal of
explaining how these sequences affect
enhancer function, however, progress in
the field has been hampered by the lack of
understanding of how enhancers activate
transcription. Enhancers are currently
viewed as entry points for transcription
factors which, once bound, can interact
with the transcription complex by looping
out intervening sequences9. More recently,
enhancer-bound transcription factors
have been shown tethered to proteins at
the promoter region that are involved in
histone modification or other types of
chromatin remodeling complexes10. Based
on these findings, the effects of insulators
on enhancers could be explained by
assuming that the insulator acts as a barrier that inhibits the processivity of a signal
traveling from the enhancer to the promoter, without inactivating either. The
leitmotiv of this scenario, sometimes
dubbed the ‘transcriptional model’, is that
the insulator acts as a promoter decoy,
confusing the enhancer-bound transcription factor into interacting with the insulator instead of the transcription
complex11. This would result in the insulator trapping the enhancer, which then
would not be able to interact with the promoter, thus precluding transcriptional
activation. Another explanation, termed
the ‘structural model’ envisions insulators
nature structural biology • volume 8 number 3 • march 2001
© 2001 Nature Publishing Group http://structbio.nature.com
news and views
a
© 2001 Nature Publishing Group http://structbio.nature.com
b
c
d
Fig. 1 Effects of insulators on enhancer–promoter interactions. The diagrams represent a generic
gene, depicted in blue, with a series of proteins forming the transcription complex present in the
promoter region. Two enhancers, named Enhancer1 and Enhancer2, serve as binding sites for
transcription factors depicted in pink and purple, respectively. An insulator is shown in brown with
two protein components represented as blue and green ovals. Green arrows indicate transcriptional activation by a specific enhancer. a, An insulator present between Enhancer1 and
Enhancer2 affects the ability of the former to activate transcription (indicated by a red X) but not
the latter. b, When an additional insulator is present between the two enhancers, the dual insulator cannot interfere with transcription activation by Enhancer1. c, The simple duplication of insulators is not responsible for their loss of function, since when Enhancer1 is present between the
tandemly repeated insulator it is unable to activate transcription. d, In a more complex situation, a
second reporter gene is present downstream from the first one. Transcription of this gene can also
be activated by Enhancer1. When tandemly repeated insulators are present and an additional
insulator is present between the two genes, both genes are transcribed.
as sequences that organize the chromatin
fiber within the nuclear space, creating
transcriptionally independent domains12
by providing chromatin with anchoring
points able to interact with some kind of
fixed substrate, perhaps the nuclear
matrix (see below). A way of explaining
both the effect of insulators on
enhancer–promoter interactions and their
ability to buffer transgenes from position
effects is to assume that the barrier is a
consequence of the involvement of insulators in the establishment of domains of
higher-order chromatin structure within
the nucleus. The formation of these
domains would then be the primary function of insulators, and their effect on
enhancer–promoter interactions would be
a secondary consequence of the establishment of these domains. Recent articles by
Cai and Shen2 and by Muravyova et al.3
seem to support this view.
Effects of insulators
Cai and Shen2 used an elegant series of
transgenic assays to assess the function of
different combinations of enhancers and
gypsy insulators in Drosophila embryos.
This insulator has been shown to contain
a series of binding sites for a zinc finger
protein called Suppressor of Hairy-wing
[Su(Hw)], which in turn interacts with a
second protein named Modifier of mdg4
[Mod(mdg4)].
When
positioned
between the zerknullt enhancer VRE
(Enhancer 1 in Fig. 1) and the evenskipped E2 enhancer (Enhancer 2 in Fig.
1), the gypsy insulator exhibited the
expected behavior, blocking only the promoter-distal VRE enhancer, while the
promoter proximal E2 enhancer activated the eve-lacZ reporter gene (Fig. 1a). A
striking result was obtained when a direct
tandem repeat of insulators was used
instead of a single copy. Not only was the
insulating effect not reinforced, it was
actually abolished, both enhancers being
able to activate transcription (Fig. 1b).
Control experiments involving other
enhancers demonstrated that the loss of
insulator activity is independent of the
enhancer studied. Also, the distance
between insulators can range from 50 to
170 base pairs with no impact on the
result. Interestingly, when flanked by a
pair of gypsy insulators, the activity of
nature structural biology • volume 8 number 3 • march 2001
the VRE enhancer is more efficiently
blocked (Fig. 1c).
Similar observations were made by
Muravyova et al.3 in their companion
paper. Using yellow and white as reporter
genes, they show that a tandem repeat of
gypsy insulators does not have any insulating properties (Fig. 1b). Indeed, the
authors note that enhancer–promoter
communication in this configuration is
improved. This holds true even when a
1.5 kilo base pair spacer was present
between the two insulators, ruling out the
trivial possibility of some steric hindrance
between insulators. These authors contribute an additional important result.
One of the constructs tested carried two
copies of the gypsy insulator between the
enhancers and the yellow gene, and an
additional copy between yellow and white.
In such a configuration both genes are
fully active (Fig. 1d).
Role of insulators in chromatin
domain organization
It is difficult to reconcile these observations with a transcriptional insulator
model. For example, a reasonable prediction from the decoy model is that two
insulators would be better than one.
That is, a tandem repeat of insulators
would reinforce the trapping of the
enhancer. Similarly, if insulators were
entry points for chromatin modifying
enzymatic complexes, the presence of
two insulators should result in a doubling of the efficiency with which chromatin remodeling complexes are
brought to the promoter region. Rather,
the results of both papers are more consistent with the structural model. In such
a model, two insulators physically interact and promote the looping of the intervening sequence. As a consequence, the
distance between enhancer and promoter is reduced, and transcription activation then takes place (Fig. 2). Similarly,
an enhancer flanked by gypsy insulators
may be looped out, resulting in inhibition of transcription. This demonstrates
that the mere interaction between
repeated copies of the insulator is not
responsible for the inability to interfere
with enhancer–promoter communication, but rather the special chromatin
structure generated by the interaction of
neighboring insulators.
Questions that remain
The ability of insulators to mediate the
formation of loops in the chromatin fiber
agrees with their possible role in the
establishment of domains of chromatin
193
© 2001 Nature Publishing Group http://structbio.nature.com
© 2001 Nature Publishing Group http://structbio.nature.com
news and views
Fig. 2 Formation of loop domains by an insulator. The diagrams show
the chromatin fiber, with the histone octamers shown in gray and the
DNA in yellow. As in Fig. 1, the coding regions of two putative genes
are shown in blue green, the transcription complex at the promoter is
shown by a series of multicolored ovals and enhancer-bound transcription factors are depicted in pink and purple. The insulator DNA is
shown in brown and two insulator proteins are shown in blue and
green. a, Interactions between insulator proteins result in the formation of a loop within the chromatin fiber, giving rise to a domain of
higher-order chromatin organization. The enhancer present outside
of this domain cannot activate transcription of the gene present
inside of the domain (indicated by a red X). b, The presence of
tandemly repeated insulators creates a small loop. An assumption of
the experimental results is that the formation of this loop between
adjacent insulators precludes interactions with other insulators. The
enhancer is then present within the same domain as the gene and
transcription activation takes place (green arrows).
organization. These type of loops have
been observed experimentally13, and
appear to be attached to a proteinaceous
backbone by sequences known as SARs
or MARs (scaffold and matrix attachment regions)14. Interestingly, the gypsy
insulator has been shown to display
SAR/MAR activity15. The structure of the
loops generated by insulators, the
dynamics of their assembly and the
mechanism of their effects on the function of enhancers are yet to be understood. One intriguing possibility is that
an enhancer/promoter pair has to reside
within the same loop to interact (Fig. 2a).
Two tandemly repeated insulators (Fig.
2b) may have a tendency to preferentially
interact with each other at the exclusion
of other insulators because of their physical proximity. The ability of the enhancer
to activate a second distal gene, bypassing
a third insulator, may be due to the
inability of the third insulator to find an
interacting partner. Conversely, if the
third insulator is able to interact with the
194
b
a
paired one, the absence of insulator function could be explained by the proximity
of the enhancer and the promoter region
of the second gene, bypassing the
requirement to reside within the same
loop. These issues will need to be
addressed before we have a clear picture
of how insulators work.
The ability of insulators to confer transgenes position-independent expression
makes these sequences extremely interesting from a practical point of view, as they
can be used to construct transgenic animals and for gene therapy purposes, to
ensure that genes are expressed at the
proper levels and in the appropriate tissues. But equally enticing is the prospect
of understanding how the DNA is organized in the nucleus. For too long the
eukaryotic nucleus has been viewed as a
black box where the cell carries the DNA.
It is now time to unravel the mysteries of
nuclear organization and bring this
organelle to the same level of understanding as the rest of the cell.
Fabien Mongelard and Victor G. Corces are
in the Department of Biology, The Johns
Hopkins University, 3400 N. Charles St.,
Baltimore, Maryland 21218, USA.
Correspondense should be addressed to
V.G.C. email [email protected].
1. Bell, A.C., West, A.G. & Felsenfeld, G. Science 291,
447–450 (2001).
2. Cai, H.N. & Shen, P. Science 291, 493–495 (2001).
3. Muravyova, E. et al. Science 291, 495–498 (2001).
4. Mihaly, J. et al. Cell Mol. Life Sci. 54, 60–70 (1998).
5. Zhou, J., Barolo, S., Szymanski, P. & Levine, M.
Genes Dev. 10, 3195–3201 (1996).
6. Hagstrom, K., Muller, M. & Schedl, P. Genes Dev. 10,
3202–3215 (1996).
7. Hark, A.T. et al. Nature 405, 486–489 (2000).
8. Bell, A.C. & Felsenfeld, G. Nature 405, 482–485 (2000).
9. Hampsey, M. & Reinberg, D. Curr. Opin. Genet. Dev.
9, 132–139 (1999).
10. Cheung, W.L., Briggs, S.D. & Allis, C.D. Curr. Opin.
Cell Biol. 12, 326–333 (2000).
11. Geyer, P.K. Curr. Opin. Genet. Dev. 7, 242–248
(1997).
12. Gerasimova, T.I., Byrd, K. & Corces, V.G. Mol. Cell 6,
1025–1035 (2000).
13. Mahy, N.L., Bickmore, W.A., Tumbar, T. & Belmont,
A.S. In Chromatin structure and gene expression,
2nd edition (eds Elgin, S.C.R. & Workman, J.L.)
300–321 (Oxford University Press, Oxford; 2000).
14. Laemmli, U.K., Kas, E., Poljak, L. & Adachi, Y. Curr
Opin. Genet. Dev. 2, 275–285 (1992).
15. Nabirochkin, S., Ossokina, M. & Heidmann, T. J. Biol.
Chem. 273, 2473–2479 (1998).
nature structural biology • volume 8 number 3 • march 2001